Opportunities and challenges for big data analytics in US higher education

Abstract

Business leaders around the world are using emerging technologies to capitalize on data, to create business value and to compete effectively in a digitally driven world. They rely on data analytics to accelerate time to insight and to gain a better understanding of their customers’ needs and wants. However, big data and data analytics solutions in higher education are new topics. There has been limited progress in accumulating the extremely rich data that flow through higher education systems for the purpose of acquiring usable information for students, instructors, administrators and the public. The key objective of this article is to propose a conceptual model for the successful implementation of analytics in higher education. The article also examines some of the potential benefits of big data and analytics as applied to the world of higher education and explores implementation challenges that can be expected. Furthermore, the study reviews key attributes of successful analytics platforms and illustrates some of the routes that might be taken to implement these technologies in education. Finally, it highlights the successful implementation of analytics solutions in several universities.

Keywords

Adaptive learning big data enrolment management predictive analytics student performance management targeted student advising

Corporations around the world are slowly beginning to incorporate big data analytics into their business models and are using it for more educated decision-making. It is expected that there will be a rapid proliferation of enterprises using business intelligence (BI) and analytics to predict the future with an acceptable level of reliability. It is also expected that data analytics will become a critical core competency for professionals of all types (Eiloart, 2017). Three major factors that have allowed for the rise of big data and big analytics have been increased computing power, large volumes of data and hardware and software innovation (Minelli et al., 2013). Although the current users of big data and big analytics are primarily large corporations, there are numerous additional industries and organizations in which a complex data system could advantageously assist decision makers (Bayrak, 2015).

One sector that big data and big analytics have yet to penetrate is that of higher education. Although there are an estimated 20.2 million students currently attending some form of higher education, there has been limited progress in accumulating and analysing data that flow through the education system (Hussar and Bailey, 2013). That said, big data can be integrated into several parts of the education system and can lead to greater student success (as measured by retention rates and knowledge acquisition) and donor giving (as measured by total annual donations) (Burroughs, 2016; Ekowo and Palmer, 2016; Hardee, 2016; Yanosky and Arroway, 2015). Therefore, it is important to explore the opportunities and challenges associated with implementing big data and big analytics in the higher education system and presenting the findings for educators, administrators and policymakers to consider.

The evolution of big data and analytics

Data are growing faster than ever before. According to Gartner Research, the volume of data will grow by 800% over the next 5 years (Groenfeldt, 2012 ). Every second, we create new data. There has been massive growth in video and photo data, triggered by staggering usage of smartphones – all packed with sensors capable of collecting a wide variety of data. There are roughly 50 billion smart connected devices in the world, and messages, updates and images posted on social networks have all contributed to the data explosion (Marr, 2015). Similarly, growing numbers of companies are collecting data from their customers. Walmart collects more than 2.5 quadrillion bytes of data every hour from its customer transactions (DeZyre, 2017). According to McAfee and Brynjolfsson, in 2012, about 2.5 billion gigabytes of data were being created each day, and that number is doubling every 40 months or so. By 2020, about 1.7 megabytes of new information will be created every second for every human being on the planet. By then, our accumulated universe of data will have reached 44 trillion gigabytes (McAfee and Brynjolfsson, 2012).

Big data defined

The term ‘big data’ was coined in mid-1990s and is defined as collections of data so large, complex and dynamic that they exceed the processing capacity of the conventional database architectures of organizations (Weiss and Indurkhya, 1998). According to Gartner, the world’s leading information technology research and advisory company, big data is comprised of high-volume, high-velocity and high-variety data (the ‘3 Vs’, as shown in Figure 1 – see, for example, https://www.gartner.com/it-glossary/big-data) (Gewirtz, 2016). These data sets are too large to be handled easily and flow in and out with excessive speed, making them difficult to analyse and, finally, the range and type of data sources are too great to assimilate (Diebold, 2012).

Figure 1.

The three Vs of big data.

The typical organization is therefore challenged in managing big data effectively, as it simply does not fit into the strictures of current database architectures. At the same time, big data draws from multiple sources and transactions and contains valuable patterns and information.

The act of gathering and storing large amounts of data for eventual analysis is not new. Since the 1950s, businesses have been using basic analytics to uncover hidden patterns and trends, show changes over time and confirm or challenge theories (Asllani, 2015). As enterprises amass broader pools of data in big data platforms, they have increased opportunities to mine those data for predictive insights. As they cannot, typically, manage the data effectively with their current database architecture, they need to seek alternative ways to process big data (Bayrak, 2015). A well-defined data management strategy is essential for the successful use of big data in corporations (Bughin, 2016). Data and analytics are playing increasingly important roles in improving competitive advantage (Taylor, 2012), and corporations see big data and the ability to analyse it as an important driver of innovation and a significant source of value creation (Tan et al., 2015).

The rise of analytics

Analytics, in the form of BI, is defined as a set of technologies, processes and tools that use data to predict likely behaviour by individuals, machinery or other entities (Mashingaidze and Backhouse, 2017). If the right type of analytics is used, big data can deliver richer insights and uncover hidden patterns and relationships. More data could translate into more possibilities for a business, but only if their real meaning can be ascertained (Minelli et al., 2013).

The new benefits that modern data analytics brings to the table are speed and efficiency. The ability to work faster – and stay agile – gives organizations a new competitive edge (Bayrak, 2015). Cloud computing technology (CCT) has emerged as the preferred technology for fulfilling the infrastructure and software needs of an enterprise via the Internet (Attaran, 2017). A recent study by the McKinsey Global Institute indicates that the pace of change is accelerating and the analytics revolution is gaining momentum, fuelled by advances in data collection and computational power. Widespread access to the cloud, insightful data visualizations, interactive business dashboards and the rise of self-service analytics have made the technology available and affordable for businesses of all sizes. Suddenly, advanced analytics is not just for the analysts (Henke et al., 2016).

Today’s analytics landscape

The past few years have seen an explosion in the business use of analytics. Corporations are using analytical tools, including BI, dashboards and data mining to gain a better understanding of their present customers and to identify potential customers and their needs. With the help of new tools, enterprises can leverage big data analytics to drive a host of business objectives, from streamlining operations to improving customer relations (Henke et al., 2016). In fact, big data analytics is set to transform virtually every business activity, bringing opportunities for enhanced customer service, optimized production levels, superior capacity planning, reduced repair and maintenance costs and improved working capital utilization (Bughin, 2016). According to a 2016 Forester study, the top three tangible analytics benefits are increased margin, profitability and increased gross sales (Evelson and Bennett, 2015). Analytics is commonly used in the areas of, for example, finance, marketing, human resources, healthcare and government policymaking (Zwilling, 2016). Several research studies have documented the advantages and widespread applications of analytics tools in corporations around the world (Eckerson, 2016; Evelson and Bennett, 2015; Gaitho, 2017; Henke et al., 2016; Lebied, 2017; Minelli et al., 2013; Roy, 2011).

Traditional reporting-based BI platforms are not designed to handle the exponential growth of the sources, volume and complexity of data. The traditional platforms enforce strict data and report governance, allowing access only to specialized reporting groups. In contrast, the modern approach views data governance as an important step in creating self-service analytics. Modern BI platforms support organizational needs for greater accessibility, agility and analytical insight from a diverse range of data sources. Moreover, while traditional systems could take months to implement, the modern approach takes as little as a few hours. Latency is no longer tolerated (Henke et al., 2016). A 2015 study by Gartner identified a shift of focus from IT-led reporting to business-led self-service analytics (Gartner, 2015). According to that study, many corporations have augmented their traditional BI platforms with more agile solutions to improve their core operations or launch entirely new business models. The modern BI platform adopted by innovative companies aims to democratize analytics through self-service capabilities such as ease of use, agility and flexibility (Table 1) (Gartner, 2015).

Table 1.

Traditional versus modern analytics.

Analytic workflow component	Traditional platform	Modern platform
Data ingestion and preparation	IT-produced	IT-enabled
Content authoring	Primarily IT staff-limited usage	Business users – comprehensive governance
Analysis	Predefined, ad hoc reporting, based on predefined model	Free-form exploration
Insight delivery	Distribution and notifications via scheduled reports or portal	Sharing and collaboration, storytelling, open APIs

Categories of analytics

Analytics is constantly evolving: It has changed dramatically over the years and is advancing rapidly today. According to Davenport and Dyche (2013), the most popular categories of analytics are descriptive, predictive and prescriptive, as shown in Figure 2. These categories build on each other and enable enterprise to make faster and smarter decisions. As organizations evolve, they move from their historical focus on ‘what’ and ‘why’ to more predictive and prescriptive analysis (Bayrak, 2015).

Figure 2.

Analytics – The key categories.

Descriptive analytics is the simplest of the three categories. It allows big data to be condensed into smaller, more useful nuggets of information. Its purpose is to summarize what happened in the past and to uncover patterns that may offer insights into business performance, so enabling users to monitor and manage their business processes more effectively (Lustig et al., 2010). In descriptive analytics, data modelling, reporting, visualization and regression are used to collect and store data efficiently, to create reports and present information and to identify trends in the data.

Predictive analytics analyses current and historical data to provide insights into what will happen and why it will happen, with an acceptable level of reliability (Abbott, 2014). It involves the use of a variety of models and techniques to project future conditions and situations (Gandomi and Heidar, 2015). It does not predict one possible future, but rather multiple futures based on the decision-maker’s actions. Statistical analysis, data mining, textual analysis, media mining, forecasting and predictive modelling are used to identify the probabilities of potential outcomes and/or the likely results of specific operations (Siegel, 2016). Predictive analytics can help businesses with a wide range of problems, and companies are using it to analyse historical data and facts to improve their understanding of clients’ needs, market potential, products, suppliers and partners and to identify potential risks and opportunities (Lebied, 2017).

Finally, the emerging technology of prescriptive analytics goes beyond the descriptive and predictive models and shows the likely outcome of each decision. It goes a step further into the future and attempts to identify what should be done and why. Prescriptive analytics employs techniques such as decision modelling, simulation and optimization to ascertain actions the organization could take to achieve the desired outcome (Lustig et al., 2010). The aim is to evaluate the effect of future decisions and to present the best course of action to take in order to adjust decisions before they are actually made (Basu, 2013). This is the most valuable category of analytics and usually results in rules and recommendations for next steps. A 2017 survey by Intel suggested that by 2020, 40% of new investment in analytics tools would be in predictive and prescriptive analytics (Intel, 2017).

Business applications of analytics

BI and analytics are more than methods of gathering and analysing data. They are about adopting the mindset of an experimenter – a willingness to let data guide a company’s decision-making process.

Organizations of all sizes are using analytics to support business core functions, such as marketing, merchandising, sales and risk management. From banking to manufacturing, from retail to healthcare, data analytics is used to make breakthrough discoveries, deliver better services and enrich the customer experience. According to recent studies, corporations are using analytics to gain various business benefits, including new revenue opportunities, improved operational efficiency, better customer service, more effective marketing and competitive advantages over rivals (Davenport and Dyche, 2013; Gaitho, 2017; Henke et al., 2016; Kalakota, 2014; Lebied, 2017; Stedman, 2017; Siegel, 2016).

Analytics is used to predict:

stock prices, risk, delinquencies, accidents, health problems, hospital admissions, malfunctions, oil flow, electricity outages, sales, donations, clicks, cancellations, fraud, tax evasion, crime, approvals for government benefits, thoughts, intention, answers, opinions, lies, grades, dropouts, friendship, romance, pregnancy, divorce, jobs, quitting, wins, votes, and more. (Siegel, 2016: 25)

The banking industry and financial institutions are using analytics to analyse the probabilities of risk and default and to prevent fraud. For example, PayPal is using predictive data analysis in its efforts to shield its users from fraud and to halt fraudulent transactions before they are processed (Burns, 2016). The healthcare industry has used analytics to process large amounts of information quickly and so to provide lifesaving diagnoses or treatment options more rapidly. It is also used in efforts to improve patient care and to identify potential health risks in advance (Stedman, 2017). Logistics companies (UPS, DHL and FedEx) have used analytics to reduce costs and improve their operational efficiency (Rosenbush and Stevens, 2015). The retail industry employs data analytics to predict customer responses or potential purchases and for upselling and cross-selling opportunities (Gaitho, 2017). Target used predictive analytics to predict from shopping behaviour which customers had the most probability of rapidly becoming pregnant (Salleh, 2013). Data analytics is also used in self-driving cars and robots (DeAngelis, 2015). As a final example, governments are using pattern recognition in images and videos for enhanced security and threat detection (Siegel, 2016; Zwilling, 2016). Other applications of analytics in the government sector include traffic control, route planning, intelligent transport systems and congestion management (Gaitho, 2017).

Overall, a recent study by the Business Application Research Center (BARC) found that organizations using data analytics reported an 8% increase in revenue and a 10% reduction in costs. Other reported benefits were better strategic decisions, better understanding of customers and improved control of operating processes (Bange et al., 2015).

The evolution of analytics in higher education

Research findings

A national review of student attainment in US higher education conducted by Indiana University and supported by a grant from the Lumina Foundation paints a bleak picture: The researchers found that only 53% of students were completing their post-secondary degree within 6 years (Shapiro et al., 2012). This report focuses on the 6-year outcomes for students who began post-secondary education in fall 2009 through spring 2012. The report found an accelerating decline in overall completion rates and a decline in completion rates across age group and enrolment intensities.

The main reason for such a decline is not necessarily lack of academic preparation. A recent analysis of 55 US colleges (see Scott, 2016; Zinshteyn, 2016) showed that

45% of college students never finished their education;

more than 40% of students who left these institutions had a grade point average (GPA) of over 3.0;

many students who had grades with a ‘B’ average or higher were not coming back after their first year; and

75% of dropouts left their studies with at least a 2.0 GPA.

These statistics were a wake-up call for many universities because they generally had not considered students at risk of dropping out unless they had a GPA of 3.0 or lower. Many students who were in the middle of the performance range and who needed help were being missed by these universities. Faced with such bleak statistics, colleges can tap big data and predictive analytics to forecast students’ success or failure and to help them to stay in school. With access to more data than ever before, analytics can become the foundation for taking action. University system officials believe predictive analytics can help to increase graduation rates by enabling educators to intervene with struggling students before failure becomes inevitable: They hope that predictive analytics can help identify pressure points that are leading students to drop out (Wells, 2016). According to Ekowo and Palmer (2016), universities are employing predictive analytics to identify those students who are most in need of advice, to develop adaptive learning courseware and to manage enrolment. Predictive analytics tools are also being used in other ways in higher education institutions, as summarized in a study by Yanosky and Arroway (2015).

Current state of big data in higher education

Big data and big analytics in higher education are relatively new topics, and there has not been significant development at any particular university or in any particular state (Daniel, 2014). A recent survey found that only 41% of the colleges surveyed used data in their decision-making and less than half of the responding schools said they were engaged in predictive analytics (Burroughs, 2016). Another survey, conducted by Educause in 2015, found only 47% of respondents identified institutional analytics as a major priority and only half as many again described learning analytics as a priority (Yanosky and Arroway, 2015). That said, one group, western interstate commission for higher education (WICHE) cooperative for educational technologies, is currently collecting unidentifiable student data from a total of 17 partnering schools to be used in analysis in the near future (Wagner and Hartman, 2016).

Recently, some colleges and universities have been using analytics for multiple purposes and for various phases of students’ academic careers (Ekowo and Palmer, 2016; Yanosky and Arroway, 2015). In a later section of this article (see the ‘Deployment of analytics in higher education’ section), we cover applications of analytics in colleges and universities around the United States.

A conceptual model for implementing analytics in higher education

Before this new technology can really be of benefit to universities, however, there has to be a fundamental shift in thinking. Analytics needs to be repositioned in the mindset of professionals working in the education sector (Ekowo and Palmer, 2017). Analytics is constantly evolving, has changed dramatically over the years and is still advancing rapidly today.

Possible analytics support for institutions during the student lifecycle

Higher education institutions can leverage analytics to transform many activities, including enrolment, student support, alumni engagement, financial aid administration and other learning and operational functions. To begin this analysis, it is helpful to consider the engagement with students from a lifecycle perspective. In the initial, pre-student stage, institutions engage with prospective students in various ways, from assisting primary and secondary education in developing educational processes to evaluating individual students for potential acceptance for a higher education programme. At the next stage of the lifecycle, the student stage, interactions with students while they are pursuing their degrees are encapsulated. Finally, there is the post-student stage, when the student becomes an alumnus of the higher education programme and may engage with the institution as a source of information about the efficacy of its programmes, advise on curricular and programme development, provide financial support and/or assist in recruiting future students for the institution. With these stages in mind, analytics can play a unique role. Figure 3 illustrates our proposed model of the use of analytics to improve efficiency in higher education institutions. The following subsections summarize the model.

Figure 3.

Using analytics to improve efficiency in higher education institutions.

Pre-student stage: Example of enrolment management (dealing with potential and incoming students)

Big data and big analytics can be used to help institutions make consistent decisions in the student admission process. At present, enrolment management departments in universities focus on a core set of data, often specific to each campus, to make decisions about which applicants to enrol. Typically, the data used include standardized test scores (SAT/ACT), high-school GPA, high-school course patterns, demographic data and specialized data such as ‘legacy’ connections. While these data have been used for years, the low rate of degree completion noted above does raise questions about how successful their use has been.

Through big data analytics, it is possible to use more of the considerable quantity of student data and statistics that campuses already have in their disparate silos to make more informed decisions. Predicting how a given student profile will perform can be done much more accurately as the volume of data from the current application process is augmented by the wealth of information in the various systems. In addition to the data already collected by the enrolment management system, information can be added to the analysis such as aggregate student performance across the range of courses (tied to standardized test performance and/or high-school GPA), the likelihood of being an on-campus or off-campus resident (based on home address), the likelihood of involvement in student organizations and/or of participating in university functions (based on high-school activities) and the likely use of campus resources ranging from the library to tutoring or advisory services (tied back to high-school activities) – all to make a better estimate of the likely performance at the university by a given applicant. Of course, these data could also be used to determine what kind of support a particular student would be likely to need if accepted.

As more and more data are collected on students, the information will begin to reveal certain trends with respect to different types of students. Ultimately, there will be enough data collected to allow prediction of a student’s potential via a mathematical calculation, given all the inputs of ethnicity, residence, SAT score, college entrance essays, high-school rank and GPA.

Further, the potential for this type of big data analytics is growing. It is already possible also to make effective use of qualitative data. For example, natural language processing technology is now available for reading essays automatically and grading them without the help of an administrator, opening the possibility of automating even the application essay process (Adams, 2014). This type of analytics automation is just one of the many ways that universities can streamline their processes and add to the collection of information on student profiles. Eventually, there will be little need for the traditional enrolment management function and a computer program will be able to predict a student’s capabilities and will do so in a way that is likely to be more accurate than the prediction of a human being. In addition to better decisions for applicants, this could have immense cost-saving implications if implemented correctly.

Student stage: Example of student performance management (dealing with current students)

More than 30 million people in the United States have earned some college credits but no degree (Shapiro et al., 2014). As we focus in on how big data and big analytics can be used by key stakeholders to increase student success, the overarching goal is to address student retention rates, time to degree completion, information retention and career preparedness. Big data analytics can assist in achieving these goals in various ways, including collecting data on student performance, identifying effective teaching methods and implementing predictive analytics based on performance.

Professors collect a plethora of data on students, such as homework scores, test scores, classroom participation and attendance, in determining the overall performance of a student in a given class. According to Daniel, ‘…Big Data analytics could be applied to examine student entry on a course assessment, discussion board entries, blog entries or wiki activity, which could generate thousands of transactions per student per course’ (Daniel, 2014: 910). Most of the data stay with the professor, and the student’s overall grade is reported into a student database. Imagine, however, if all of the student’s work leading up to the final grade were also reported into the system and a profile for each student were built. For some larger schools, this would create millions of transactions over the course of a single year, which it would be impossible for any department to manage.

That said, a complex analytics system would be able to capture, analyse and generate meaningful data correlations and patterns. This type of system could then make correlations, such as that between the number of absences from class and the student’s final grade. If there were a significant correlation, then the system could be set up to identify students at risk from frequently missing class. This type of system could also analyse trends in individual students across time. If, for example, a student were performing poorly in writing assignments in different classes, the system could notify the student and campus writing centre. Further, it could recommend additional coursework for particular students based on their results across classes. Using the data collected, administrators could identify those areas in which they were above the national average and those in which they were below it, making curriculum adjustments accordingly.

Colleges could also use a data analytics system to identify which teaching methods lead to better understanding and more long-term retention. One way to implement this is to have the same professor teach a class in two or three different ways. Perhaps one class would be set up with primarily student projects, presentations and no exams, while another would consist primarily of exams and essays. The effectiveness of teaching methods can then be tested by giving students an examination at the beginning of the term to test their baseline knowledge and giving them the same examination at the end of the term to test their cumulative retention of concepts. It would also be possible to look more deeply into the type of learning improvements required. For example, in the quantitative area, more data would allow administrators to determine whether a learning problem was in understanding and framing the question, in the actual computations or in the analysis afterwards and making sense of the computations. Rather than a generalized finding of a need for better quantitative skills – a finding that does not really offer a clear prescription – a more targeted finding would offer a clearer and more effective course of remediation. After a few years of testing, a well-developed analytics system can show trend analysis and demonstrate which teaching methods are most effective in promoting overall student retention.

Post-student stage: Example of university advancement (analytics for donor relations and federal funding)

Colleges and universities in the United States can greatly benefit from the implementation of an analytics system in terms of gaining more federal funding and improving donor relations. An environment conducive to such usage has been developing. For example, while in office, President Obama worked on a strategy to make colleges more affordable for the middle class by promoting new policies that specified performance-based funding (Nisar, 2015). Essentially, the proposal was to identify specific factors, like time to degree and affordability, and then to allocate federal funds to the schools that provided the best balance between low cost and ability to graduate students. In turn, this would incentivize universities to find ways to keep their costs low and to graduate students in a timely manner. In response, it is likely that some universities will adopt analytics systems that do a better job of tracking student progress, measuring the most effective teaching strategies and enrolling students whose predicted potential is the greatest, as well as reporting these results to government funding agencies.

Big data analytics systems can also facilitate stronger relations between the university and potential donors. For example, an analytics system can track donor information, such as residence, income, ethnicity, the amount previously donated, community affiliations and other metrics to build a database with all of the information that will identify trends. Eventually, the system will be able to predict which neighbourhoods, ethnicities and income levels are most likely to donate money to the university. Administrative personnel can then focus on developing relationships with those who fall into those categories and avoid those who are not likely to donate. Ultimately, the university will benefit because it will waste less time contacting individuals who are not likely to donate and will increase its endowment fund by focusing efforts where they will do the most good.

Promises and challenges of analytics in higher education

Universities in the United States are under pressure from government, parents and students to do a better job of graduating students. Performance-based funding has increased the pressure to ensure that every student succeeds (NCSL, 2015). Besides, universities can minimize loss of revenue from tuition and fees by retaining students because it costs less to retain a student than to recruit a new one (Eduventures, 2013). In this context, analytics could be viewed as an empowering tool which helps institutions to create an enriched learning experience for the students and raise graduation rates. With access to more data and the availability of easy-to-use predictive analytics tools, more universities can promote academic success for students (Yanosky and Arroway, 2015).

However, although universities are sitting on huge amounts of useful data, these data are stored in different departments and software systems and are not used to connect the dots. Powered by a cloud computing infrastructure, which offers cost savings, scalability, agility and modernization, this situation can be transformed so that universities are making effective and coordinated use of the wealth of data they collect (Attaran et al., 2017). Until recently, much of the data were collected for accountability purposes and were merely shared with state and federal agencies that track the success of universities in graduating their students. By using predictive analytics to pull all of those data together, universities can ascertain, for instance, how often students interact with online course materials, or whether freshman students who receive a ‘C’ in certain courses are more or less likely to graduate. Prognostic analytics can be used in digital courses to identify what a student is learning and what components of a lecture plan most effectively teach them (Wells, 2016).

Likely challenges and setbacks

Although, as the above discussion has emphasized, there are very real and practical applications of big data and big analytics for higher education, they certainly will not be implemented unless a variety of challenges and potential setbacks are addressed. For example, many universities lack analytics skills and do not have the internal resources to take advantage of a wealth of data-driven insights. As a result, they outsource analytics or, more often, simply fail to leverage the information they already possess (McGuirt et al., 2015). In a 2015–2016 Higher Education Industry Outlook Survey of 102 senior higher education leaders, 41% indicated that they used data/analytics for forecasting, and 36% of colleges outsourced analytics because they lacked the necessary skills internally. Moreover, 29% had the resources to analyse data for strategic and operating decisions, and 22% said that they had sufficient data but did not incorporate it effectively in decision-making. The survey identified the key challenges facing institutions as effectively using data residing in different functions for decision-making, data quality, dealing with new types of data and adopting new or more advanced analytics techniques (McGuirt et al., 2015).

An Educause survey in 2015 identified a host of challenges in implementing analytics systems, including organizational behaviour issues such as resistance to change and a lack of vision, a lack of appropriate financial resources, a shortage of analysts (digital skills) and insufficient computing power (Yanosky and Arroway, 2015). Anderson and Ackerman-Anderson (2001) suggest that when implementing workplace change, there has to be some sort of call to action or ‘wake-up call’ in order to legitimize the change and incentivize leaders to acknowledge its benefit. Further, leaders must take an active role in ‘creating organizational vision, commitment, and capacity’ (Anderson and Ackerman-Anderson, 2001: 40–41). In other words, the university’s leaders must first understand the benefit of implementing a complex analytics system and must then show commitment to and spearhead the adoption of the new technologies. Furthermore, there has to be a certain level of capacity in terms of computing power, human capital and financial resources which must be allocated to implement the system successfully – something that can often be difficult and can create dissent. Finally, some concerns may arise from what the data may indicate. More specifically, administrators may be hesitant if they believe that the information might reveal that their students are underperforming compared to the national average, that their graduation rates are lower than those of similar universities or that certain ethnic minorities are performing below average – all of which might draw public attention and scrutiny.

Arguments against analytics in higher education

Despite the potential of using big data analytics in universities, there are some concerns with regard to possible unfairness in predicting students’ potential and the invasion of student or applicant privacy. One argument against using predictive analytics in the admissions decision process is that the system does not take into consideration specific circumstances, thus creating a type of automated stereotyping. Take, for example, the case of a student who performs very poorly in one class due to a family illness or some other serious distraction. The student will have a significantly lower GPA due to that one class and may well not be admitted. Another example is the student from a particular geographical location or ethnicity that, according to the data, tends to produce below-average performing students. That student would automatically be disadvantaged solely because the system had projected a performance consistent with the dominant trend, creating a kind of systemic adverse impact. Many would argue that this would be unfair, unethical and possibly illegal, so there must be strict limits on what can and cannot be incorporated into an analytics system.

Another major concern with implementing this type of system relates to the safeguarding of information about students and the potential invasion of privacy. Although college classes have traditionally evaluated students on the basis of their performance and behaviour, big data changes the level and scope of the analytics, and so there is a need for careful evaluation of the implications and potential impacts (Picciano, 2012). Essentially, there must be safeguards in place to ensure that individuals cannot obtain unauthorized access to the data and that the data are ‘relevant to the purposes for which they are to be used, and, to the extent necessary for those purposes’ (Watermand and Bruening, 2014: 90). Such safeguards should include data encryption and limited authorization to access the data. That said, there is undeniably an ethical dilemma as to whether the data should be collected at all, and whether or not the benefits outweigh the costs. This will ultimately be decided by a combination of decisions from students, administrators and policymakers.

Deployment of analytics in higher education

Today’s higher education institutions struggle to obtain measurable value from their BI tools because of the fragmented nature of the data, security gaps and the confusion caused by incomplete information (Ekowo and Palmer, 2016). Some universities have used data analysis for years, but their analytics expertise has been contained within small pockets of the organization. Some teams have lacked access to analytics, and data management practices have been inconsistent. To become a data-driven organization, a university needs a thoughtfully designed analytics platform that empowers everyone to make data an integral part of their day-to-day processes and decisions. It needs an analytics solution that can bring together disparate data in a governed environment that allows users from different departments to model, discover, communicate and distribute information easily (Ekowo and Palmer, 2017). In most cases, universities need to look beyond their own IT staff for assistance: A knowledgeable and experienced service provider can provide initial and ongoing assistance to increase the chance of analytics success (Burroughs, 2016).

A range of recently published research on the implementation of BI and predictive analytics has been reviewed to explore the current status, issues and challenges identified (Asllani, 2015; Ekowo and Palmer, 2017; Loshin, 2017a, 2017b; McNeill, 2014). It is argued that, before deploying analytics process in an organization, the areas in which analytics will add business value need to be identified and a scalable deployment approach must be planned. We have modified guiding practices suggested by these papers and added more to create practical implementation steps for the successful deployment of a predictive analytics process in a higher education institution. The following summarizes our recommended implementation steps.

Key factors to consider

The key attributes of a well-designed analytics system are the following:

Vision and plan. Develop a vision and plan for data use that help to steer a predictive analytics effort. Include in your plan the questions you hope to answer and the goals you aim to achieve. Explore the potential pitfalls of using student data. Make sure that data will not be used for discriminatory purposes. Be sure to include key staff in decision-making.

Scalability. Consolidate disparate data into a shared repository-based platform that provides scalable self-service solutions to all decision-makers in the organization. This facilitates the spreading of data-driven decision-making and optimization to departments and other units in the university.

User-friendly interfaces. Make analytics easy to use for everyone. All decision-makers from different segments of the university must have access to information normally dependent on complicated and sophisticated analytics tools. On-demand help and robust online guides should be built into the system to answer any questions.

Up-to-date. Avoid outdated and irrelevant analytics. Avoid restricting analytics edits. Conduct analytics directly on real-time data.

Real-time collaboration. Democratize data-driven analytics. Avoid limiting access to analytics and make sharing analytics and context simple for decision-makers. Expand the use of data throughout the university. Allow analytics users to slice the data and answer their own questions. Ensure consistency and coordination across the organization.

Quick installation, maintenance and upgrade. Analytics tools can be installed in a matter of hours or days and should be simple for the IT department to maintain and upgrade.

Reliability and security. Make sure your analytics solutions guarantee that your data are accurate, available and audited. Create a strong partnership with your IT department to ensure that data are accurate. Also make sure that your analytics solutions provide proper security options that allow users to create and publish their work securely. Finally, ensure that complete security controls and a trail of users are available at all times.

Higher education institutions can leverage analytics to drive a host of business objectives; however, finding the right analytics solution for your campus can be challenging. There is no one-size-fits-all option and no plug-and-play device. There is a wealth of new tools that leverage analytics for specific purposes, and independent standards are being developed rapidly by vendors. Several companies are providing analytics solutions in the education sector, including SAP, Tableau Software, SPSS and Rapid Insight. Table 2 provides a summary of these services.

Table 2.

Analytics solutions available to educational institutions.

Analytics service provider	Analytics solutions	Institution
Tableau Software	Tableau	California State University Systems, University of Texas at Austin, Washburn University
SAP	SAP HANA	University of Kentucky
QlicView	Olic	Excelsior College
SPSS	SPSS	Washburn University
Rapid Insight, Inc.	Rapid Insight	Sarah Lawrence College
Blackboard	Blackboard Analytics	CUW Piedmont Community College
IBM	BM Predictive Analytics software	Michigan State University
SAAS	Visual Analytics	Sinclair Community College, University of Oregon
Civitas Learning	College Scheduler	Arizona State University, University of Arizona

Note: CUW: Concordia University Wisconsin.

Case examples of success

Applications in colleges and universities

Progressive higher education institutions are implementing analytics across the student lifecycle to attract the right students, maximize student retention and graduation rates, gain more federal funding and improve donor relations. Table 3 offers examples of US universities and colleges that are already using predictive analytics to optimize key phases of the student lifecycle and to align their resources with their organizational goals. For each institution, the analytics objectives, the processes targeted to achieve those objectives and the benefits gained are summarized.

Table 3.

Applications of analytics in American colleges and universities.

Institutions	Process targeted	Objectives	Benefits gained
Michigan State University	University advancement department	Identifying potential donors and providing deep insight into an individual alum’s potential to give	Improved director and associate director productivity; improved visibility of donor patterns; improved overall user productivity; annual labour savings of US$34,434
CUW	Student performance management	Identifying at-risk students and helping them	Increased student retention rate to 82%, a 10% increase in 1 year
University of California – Santa Barbara	University advancement department	Improving visibility of who will donate and repeat donors	Saved time and money; exponentially increased yearly revenue from donors
Arizona State University	Student performance management	Improving students’ course schedules	Graduation rates climbed by 20%
University of Maryland – College Park	Student performance management	Predicting student success or failure and intrusive advising	Helped narrow achievement gaps for minority and low-income students; improved graduation rates; shortened graduation time
Mount St Mary’s University in Emmetsburg	Student performance management	Predicting which incoming freshmen were unlikely to succeed	Boosted graduation rates by helping struggling students
Georgia State University	Student performance management	Timely intervention for students with high chance of dropout	Graduation rates raised by six percentage points; eliminated achievement gaps for low-income and minority students
Johns Hopkins University	Student performance management	Flagging students who are missing assignments or skipping class	Helped increase graduation rates
University of Texas at Austin	User management and security	Timely and accurate intervention in security threats and incidence across the distributed university network	Faster insight into anomalies; improved security posture; educed organizational risk; reduced incident investigation time
Washburn University	Student performance management	Retention and graduation	Helped increase graduation rates
Sinclair Community College	Student enrolment management	Generate notification to send to individual student	25–33% increases in enrolments year over year
University of Oregon	Financial aid programme	Redesign the merit-aid programme to recruit high achievers more effectively	Enabled deeper insights into the behaviour of applicants who were accepted and offered merit aid, thus increasing the likelihood that these students would enrol; learned how much merit aid is needed in a financial aid package to make high-achieving, in-state students more likely to enrol
Delaware State University	Student enrolment management	Identify students at risk and streamline best advising practices	Increased student retention rate to 70%; strived to improve 4-year graduation rate by 4 percentage points per year

Source: Beckwith (2016), Ekowo and Palmer (2016), Hardee (2016), Felton (2016), Johnson (2016), Tissot (2017), Wells (2016), Wicom et al. (2011) and Zinshteyn (2016).

Note: CUW: Concordia University Wisconsin.

Most colleges do not have an adequate number of student advisors, and so it is not possible for them to give individual students the personalized attention they need. Predictive analytics can help colleges to pinpoint those students most in need of institutional support in two ways: Early-Alert and Program Recommender Systems (Ekowo and Palmer, 2016). Early-Alert Systems use predictive analytics to identify at-risk students. Predictive models can include high-school and college GPA, demographic data, class attendance and course-taking patterns. Recommender Systems, on the other hand, use predictive analytics to help identify courses or programmes for students. Predictive analytics have also been used by universities for enrolment management (Beckwith, 2016), to identify struggling students and streamline advising practices (Hardee, 2016), to anticipate the financial needs of incoming and returning classes and to determine whether or not a student will accept the financial aid award offered (Ekowo and Palmer, 2016).

In the following, we provide specific examples of data analytics applications in US higher education institutions.

Case studies – Targeted student advising

Temple University, a public research university in Philadelphia, uses analytics to help identify students who are at risk of struggling academically and in danger of dropping out. Between 2001 and 2014, Temple’s use of predictive analytics resulted in a 24% increase in its 4-year graduation rate, an 11% increase in its 6-year graduation rate and an increase of 12% in the proportion of students who returned for a sophomore year (Felton, 2016).

Officials at Arizona State University are also using analytics to improve students’ academic experience and have managed to boost graduation by 20%. The university is using ‘College Scheduler’ analytics software, a platform that enables students to enter personal information into a dashboard. The programme considers students’ personal and academic obligations and auto-populates the courses they have to take. The software is valuable because it prevents students from taking courses that do not count towards their majors, thus wasting their time and financial aid. College Scheduler has been shown to boost college completion rates by more than 3% (Zinshteyn, 2016).

Concordia University Wisconsin (CUW) has also successfully implemented an analytics programme to identify at-risk students and help them out. Blackboard intelligence and learning analytics solutions are used to ascertain how students are developing in their academic commitments while there are still opportunities for recovery if they are doing poorly. Student advisors focus on student performance and a variety of risk factors based on the data provided and with the help of dashboards. Advisors use dashboards to support students more efficiently. The use of data analysis has improved CUW’s retention rate by 10%: In 2016, the university had a retention rate of 72% and a year later, using analytics along with better involvement of faculty and administration, CUW reached an 82% retention rate.

Officials at the University of Maryland, College Park, analyse student data, including grades, demographics, financial aid, course schedules and enrolment status, to identify at-risk students and improve retention rates. They use predictive analytics to intervene with struggling students before it is too late. Analytics helps to identify bottlenecks and problems, such as a difficult class or other pressing issues that could lead a student to drop out (Wells, 2016). For example, one finding from this approach is that students who enrol in a course very late tend not to perform well in it. Therefore, the institution’s policy is now not to let any student enrol in a class in the last few days before it starts, although it is still possible to drop a class four days after it has started without a penalty. The university is also using a data tool called ‘Student Success Matrix’ that has been developed by the non-profit Predictive Analytics Reporting Framework. Using this tool, officials can determine whether a C grade in an introductory marketing course indicates a low chance of the student graduating in the major. They then use ‘intrusive advising’ where appropriate to the student improve grades or change major. Additionally, the university found that students who received bad grades (D–F) used Blackboard 40% less than those who received A, B or C. Instructional technology experts built a tool using Blackboard called ‘Check-My-Activity’, which enables students to compare their Blackboard activity with that of anonymous classmates who received a higher or lower grade on an assignment. This use of feedback for intervention is already producing good results. According to officials, students who use the Check-My-Activity tool are nearly three times more likely to earn at least a C grade (Wells, 2016).

Case studies – Adaptive learning

Universities use predictive analytics to develop adaptive learning courseware which modifies a student’s learning route to enhance and accelerate learning (Ekowo and Palmer, 2016). One example is provided by Colorado Technical University, a private institution that offers undergraduate and graduate degrees primarily online. In 2012, the university adopted predictive analytics to develop adaptive learning courseware. It used ‘Intellipath’ to assess what students did and did not know, and then presented information to help them meet the course learning goals quickly. Intellipath improved student engagement and retention: For example, 81% of students passed the Accounting I course, 95% completed the course and the average grade increased from C to B (Johnson, 2016).

Summary and conclusion

Higher education institutions are operating in an increasingly complex and competitive environment. This article has identified some of the challenges they face and has explored the potential of analytics to address these challenges. In addition, the article uses the three lifecycle stages of student engagement and proposes a conceptual framework for applications of analytics in higher education.

As has been shown, US colleges and universities can use big data and data analytics in a variety of ways to help them make better decisions. More specifically, big data analytics can help enrolment management personnel with their admission decisions, enable university professionals to identify students who are in need of campus resources and help universities to gain more funding through donations. That said, because only a small proportion of universities currently use data analytics, there will be a significant need for standards and best practices when more universities have introduced these types of programmes. Implementing such programmes may encounter resistance from those who do not think that the benefits justify the investment. For a smooth and effective implementation, therefore, it is essential that the university’s leaders are all committed to the initiative and ready to support its development. In addition, lines will have to be drawn with regard to how much data universities can collect and, specifically, to what uses it can be put. Although collecting data to help students succeed in classes is probably a good idea, collecting excessive data with a lack of strict usage controls might be seen as violating student rights and privacy. That said, examples of successful implementation such as those provided by CUW and the University of Maryland suggest that the effort and investment are well worthwhile.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Abbott

(2014) Applied Predictive Analytics: Principles and Techniques for the Professional Data Analyst. Hoboken, NJ: John Wiley & Sons.

Adams

(2014) Essay-grading software seen as time-saving tool. Education Week, 10 March 2014. Available at: https://www.edweek.org/ew/articles/2014/03/13/25essay-grader.h33.html (accessed 19 March 2018).

Anderson

Ackerman-Anderson

(2001) Beyond Change Management: Advanced Strategies for Today’s Transformational Leaders. San Francisco, CA: Jossey-Bass/Pfeiffer.

Asllani

(2015) Business Analytics with Management Science, Models and Methods. Upper Saddle River, NJ: Pearson Education.

Attaran

(2017) Cloud computing technology: leveraging the power of the Internet to improve business performance. International Journal of Information Technology and Management 26(1): 112–137.

Attaran

Celik

(2017) Promises and challenges of cloud computing in higher education: a practical guide for implementation. Journal of Higher Education Theory and Practice 17(6): 20–38.

Bange

Grosser

Janoschek

(2015) Big Data Use Cases: Getting Real on Data Monetization. BARC Research Study. Available at: http://barc-research.com/wp-content/uploads/2015/07/BARC_Big_Data_Use_Cases_EN_2015.pdf (accessed 19 March 2018).

Basu

(2013) Five pillars of prescriptive analytics success. Analytics Magazine, 8–12 March–April. Available at: http://analytics-magazine.org/executive-edge-five-pillars-of-prescriptive-analytics-success/ (accessed 20 March 2018).

Bayrak

(2015) A review of business analytics: a business enabler or another passing fad. Procedia – Social and Behavioral Sciences 195: 230–239.

10.

Beckwith

(2016) Data analytics rising in higher education: a look at four campus ‘data czars’ and how they’re promoting predictive analytics. University Business Magazine. Available at: https://www.universitybusiness.com/article/data-analytics-rising-higher-education (accessed 19 March 2018).

11.

Bughin

(2016) Big data, big bang? Journal of Big Data 3(2). Available at: https://journalofbigdata.springeropen.com/articles/10.1186/s40537-015-0014-3 (accessed 19 March 2018).

12.

Burns

(2016) How PayPal fights fraud with predictive data analysis. Available at: http://searchbusinessanalytics.techtarget.com/feature/How-PayPal-fights-fraud-with-predictive-data-analysis (accessed 19 March 2018).

13.

Burroughs

(2016) Survey: data and analytics in higher ed can be a one-two punch. EdTech Magazine. 8 April 2016. Available at: https://edtechmagazine.com/higher/article/2016/04/survey-data-and-analytics-higher-ed-can-be-one-two-punch (accessed 19 March).

14.

Daniel

(2014) Big data and analytics in higher education: opportunities and challenges. British Journal of Educational Technology 46(5): 904–920.

15.

Davenport

Dyche

(2013) Big data in big companies. International Institute for Analytics. Available at: http://www.datascienceassn.org/sites/default/files/Big%20Data%20in%20Big%20Companies%20-%20Tom%20Davenport.pdf (accessed 19 March 2018).

16.

DeAngelis

(2015) Predictive analytics becoming a mainstream business tool. Available at: https://www.enterrasolutions.com/blog/predictive-analytics-becoming-a-mainstream-business-tool/ (accessed 19 March 2018).

17.

DeZyre (2017) How big data analysis helped increase Walmart’s sales turnover? Available at: https://www.dezyre.com/article/how-big-data-analysis-helped-increase-walmarts-sales-turnover/109 (accessed 19 March 2018).

18.

Diebold

(2012) On the origin(s) and development of the term ‘Big Data’. PIER Working Paper 12-037. Penn Institute for Economic Research, Department of Economics, University of Pennsylvania. Available at: https://economics.sas.upenn.edu/sites/economics.sas.upenn.edu/files/12-037.pdf (accessed 19 March 2018).

19.

Eckerson

(2016) Embedded Analytics: The Future of Business Intelligence. Hingham, MA: Eckerson Group.

20.

Eduventures (2013) Predictive analytics in higher education: data-driven decision-making for the student life cycle. Available at: https://public.dhe.ibm.com/common/ssi/ecm/yt/en/ytl03202usen/YTL03202USEN.PDF (accessed 19 March 2018).

21.

Eiloart

(2017) Top trends in business intelligence for 2017. Available at: https://www.itproportal.com/features/top-trends-in-business-intelligence-for-2017/ (accessed 19 March 2018).

22.

Ekowo

Palmer

(2016) The Promise and Peril of Predictive Analytics in Higher Education. New America report. Available at: https://na-production.s3.amazonaws.com/documents/Promise-and-Peril_4.pdf (accessed 19 March 2018).

23.

Ekowo

Palmer

(2017). Predictive Analytics in Higher Education: Five Guiding Practices for Ethical Use. New America report. Available at: https://na-production.s3.amazonaws.com/documents/Predictive-Analytics-GuidingPractices.pdf (accessed 19 March 2018).

24.

Evelson

Bennett

(2015) Quantify the Tangible Business Value of BI. Forrester Research report. Available at: http://www.lavastorm.com/assets/2015-Forrester-Quantify-Tangible-Business-Value-of-BI.pdf (accessed 19 March 2018).

25.

Felton

(2016) Temple University is spending millions to get more students through college, but is there a cheaper way? The Hechinger Report, 18 May 2016. Available at: http://hechingerreport.org/temple-university-spending-millions-get-students-college-cheaper-way/ (accessed 19 March 2018).

26.

Gaitho

(2017) How applications of big data drive industries. Available at: https://www.simplilearn.com/big-data-applications-in-industries-article (accessed 19 March 2018).

27.

Gandomi

Haider

(2015) Beyond the hype: big data concepts, methods, and analytics. International Journal of Information Management 35(2): 137–144.

28.

Gartner (2015) Gartner survey shows more than 75 percent of companies are investing or planning to invest in big data in the next two years. Gartner, Press Release, 16 September 2015. Available at: http://www.gartner.com/newsroom/id/3130817 (accessed 19 March 2018).

29.

Gewirtz

(2016) ‘Volume, velocity, and variety: understanding the three V’s of big data. Available at: http://www.zdnet.com/article/volume-velocity-and-variety-understanding-the-three-vs-of-big-data/ (accessed 19 March 2018).

30.

Groenfeldt

(2012) Big data – big money says it is a paradigm buster. Forbes. Available at: https://www.forbes.com/sites/tomgroenfeldt/2012/01/06/big-data-big-money-says-it-is-a-paradigm-buster/#6120d0bfe389 (accessed 6 January 2012).

31.

Hardee

(2016) Better by the numbers. Available at: http://www.impatientoptimists.org/home/posts/2016/01/better-by-the-numbers#/h (accessed 20 March 2018).

32.

Henke

Bughin

Chui

. (2016) The Age of Analytics: Competing in a Data-Driven World. McKinsey Global Institute Report. Available at: https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/the-age-of-analytics-competing-in-a-data-driven-world (accessed 19 March 2018).

33.

Hussar

Bailey

(2013) Projections of Education Statistics to 2022. NCES 2014-051, US Department of Education, National Center for Education Statistics. Washington, DC: US Government Printing Office.

34.

Intel (2017) Guide to Getting Started with Advanced Analytics. Intel Planning Guide. Available at: https://www.intel.com/content/www/us/en/analytics/getting-started-advanced-analytics-planning-guide.html (accessed 19 March 2018).

35.

Johnson

(2016) Adaptive learning platforms: creating a path for success. Educause Review, 7 March 2016. Available at: http://er.educause.edu/articles/2016/3/adaptivelearning-platforms-creating-a-path-for-success (accessed 19 March 2018).

36.

Kalakota

(2014) A primer on predictive analytics. Available at: http://www.amberoon.com/CarpeDatumRx/bid/338658/A-Primer-on-Predictive-Analytics (accessed 19 March 2018).

37.

Lebied

(2017) Top 11 business intelligence and analytics trends for 2017. Available at: https://www.techombay.com/news-article/top-11-business-intelligence-and-analytics-trends-for-2017/ (accessed 19 March 2018).

38.

Loshin

(2017a) Five steps to build better predictive analytics applications. Available at: http://searchbusinessanalytics.techtarget.com/tip/Five-steps-to-build-better-predictive-analytics-applications (accessed 19 March 2018).

39.

Loshin

(2017b) Ten steps to start using predictive analytics algorithms effectively. Available at: http://searchbusinessanalytics.techtarget.com/tip/Ten-steps-to-start-using-predictive-analytics-algorithms-effectively (accessed 19 March 2018).

40.

Lustig

Dietrich

Johnson

. (2010) The analytics journey. Analytics Magazine, 11–18 November/December. Available at: http://analytics-magazine.org/the-analytics-journey/ (accessed 20 March 2018).

41.

Marr

(2015) Big data: 20 mind-boggling facts everyone must read.’ Forbes, 30 September 2015. Available at: https://www.forbes.com/sites/bernardmarr/2015/09/30/big-data-20-mind-boggling-facts-everyone-must-read/#56a1b34e17b1 (accessed 19 March 2018).

42.

Mashingaidze

Backhouse

(2017) The relationships between definitions of big data, business intelligence and business analytics: a literature review. International Journal of Business Information Systems 26(4): 488–505.

43.

McAfee

Brynjolfsson

(2012) Big data: the management revolution. Harvard Business Review, October. Available at: https://hbr.org/2012/10/big-data-the-management-revolution (accessed 19 March 2018).

44.

McGuirt

Gagnon

Meyer

(2015) Embracing Technology: 2015–2016 Higher Education Industry Outlook Survey. KPMG report. Available at: http://www.kpmg-institutes.com/content/dam/kpmg/governmentinstitute/pdf/2015/he-outlook-2016.pdf (accessed 19 March 2018).

45.

McNeill

(2014) Analytics in Healthcare and the Life Sciences: Strategies, Implementation Methods and Best Practices. Upper Saddle River, NJ: Pearson.

46.

Minelli

Chambers

Dhiraj

(2013) Big Data, Big Analytics: Emerging Business Intelligence and Analytic Trends for Today’s Businesses. Hoboken, NJ: John Wiley & Sons.

47.

NCSL (2015) Performance-based funding for higher education. In: National Conference of State Legislators, 31 July 2015. Available at: http://www.ncsl.org/research/education/performance-funding.aspx (accessed 19 March 2018).

48.

Nisar

(2015) Higher education governance and performance based funding as an ecology of games. Higher Education 69(2): 289–302.

49.

Picciano

(2012) The evolution of big data and learning analytics in American higher education. Journal of Asynchronous Learning Networks 16(3): 9–20.

50.

Rosenbush

Stevens

(2015) At UPS the algorithm is the driver. The Wall Street Journal, 16 February 2015. Available at: https://www.wsj.com/articles/at-ups-the-algorithm-is-the-driver-1424136536 (accessed 19 March 2018).

51.

Roy

(2011) Analyse this! Analytics explained and applied. Available at: https://www.capgemini.com/gb-en/2011/05/analyse-this-analytics-explained-and-applied/ (accessed 19 March 2018).

52.

Salleh

(2013) Applying predictive analytics in enterprise decision making. Available at: http://www.lumina.com/blog/applying-predictive-analytics-in-enterprise-decision-making (accessed 19 March 2018).

53.

Scott

(2016) Colleges tap big data to help students in school. Available at: https://www.marketplace.org/2016/07/20/education/tapping-big-data-help-college-students-succeed (accessed 19 March 2018).

54.

Shapiro

Dundar

Chen

. (2012) Completing College: A National View of Student Attainment Rates. Signature Report No 4. Herndon, VA: National Student Clearinghouse Research Center. Available at: https://offices.depaul.edu/enrollment-management-marketing/student-retention/Documents/NSC_Signature_Report_4.pdf (accessed 19 March 2018).

55.

Shapiro

Dundar

Yuan

. (2014) Some College, No Degree: A National View of Students with Some College Enrollment, but No Completion. Signature Report No 7. Herndon, VA: National Student Clearinghouse Research Center. Available at: https://nscresearchcenter.org/wp-content/uploads/NSC_Signature_Report_7.pdf (accessed 19 March 2018).

56.

Siegel

(2016) Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die. Hoboken, NJ: John Wiley & Sons.

57.

Stedman

(2017) Eyeing the future with predictive analytics can pay dividends now. Available at: http://searchbusinessanalytics.techtarget.com/ehandbook/Predictive-data-analytics-advances-businesses-ahead-of-the-game (accessed 19 March 2018).

58.

Tan

Zhan

Guojun

. (2015) Harvesting big data to enhance supply chain innovation capabilities: an analytic infrastructure based on deduction graph. International Journal of Production Economics 165: 223–233.

59.

Taylor

(2012) Crunch time for big data. Financial Times, 19 June 2012. Available at: https://www.ft.com/content/bd5a5ce2-aa57-11e1-899d-00144feabdc0 (accessed 19 March 2018).

60.

Tissot

(2017) Concordia University Wisconsin: how to achieve a record retention rate. E-Learn, February 2018: 29–32. Available at: https://elearnmagazine.com/wp-content/uploads/2018/01/EN-E-Learn-Special-Edition-Learning-Analytics.pdf (accessed 23 March 2018).

61.

Wagner

Hartman

(2016) Welcome to the era of big data and predictive analytics in higher education. Available at: http://www.sheeo.org/sites/default/files/0808-1430-plen.pdf (accessed 19 March 2018).

62.

Waterman

Bruening

(2014) Big data analytics: risks and responsibilities. International Data Privacy Law 4(2): 89–95.

63.

Weiss

Indurkhya

(1998) Predictive Data Mining: A Practical Guide. Burlington, MA: Morgan Kaufmann.

64.

Wells

(2016) Maryland universities to use data to predict student success – or failure. The Baltimore Sun, 11 June 2016. Available at: http://www.baltimoresun.com/news/maryland/education/bs-md-college-analytics-20160611-story.html (accessed 19 March 2018).

65.

Wicom

Ariyachandra

Goul

. (2011) The current state of business intelligence in academia. Communications of the Association for Information Systems 29(16): 299–312.

66.

Yanosky

Arroway

(2015) The Analytics Landscape in Higher Education. Louisville, CO: Educause Center for Analysis and Research. Available at: https://library.educause.edu/∼/media/files/library/2015/5/ers1504cl.pdf (accessed 19 March 2018).

67.

Zinshteyn

(2016) The colleges are watching. Atlantic Daily, 1 November 2016. Available at: https://www.theatlantic.com/education/archive/2016/11/the-colleges-are-watching/506129/ (accessed 20 March 2018).

68.

Zwilling

(2016) 10 ways to use analytics to supercharge your business. Huffpost – The Blog, 17 February 2016. Available at: http://www.huffingtonpost.com/marty-zwilling/10-ways-to-use-analytics_b_9254166.html (accessed 19 March 2018).